基于贝叶斯蒙特卡罗的液氨罐区安全系统动态风险分析

Safety system dynamic risk analysis of the ammonia tank by using Bayesian Monte Carlo simulation

以概率分布表示液氨罐区各安全系统的失效率,用贝叶斯蒙特卡罗对安全系统进行动态分析。首先,构建事件树;然后,用无信息的均匀分布作为各个安全系统失效率的先验分布,由结果事件发生次数得到似然函数,进而得到安全系统失效的后验分布;最后,由蒙特卡罗模拟得到事件树结果事件发生的概率值。结果表明,火灾和中毒发生的概率比先验值大,液氨罐区总体安全性能降低。报警器、自动关断系统和灭火系统失效率比先验失效率高;浓度传感器、手动关断系统和防火系统失效率比先验失效率低。后验分布的方差逐渐减小,风险评估的不确定性降低。因此,动态分析更能够反映液氨罐区安全系统的安全状态,有利于决策人员及时采取相应措施。

The paper is concerned about the authors＇ dynamic risk analysis of the ammonia tank by using Bayesian Monte Carlo Simula tion while updating the failure probability distribution of the safety system and accident-incidence probability in each end state. First of all, we use an event tree to denote the potential accident scenarios. Then, no-informative prior distribution method has been used as the prior failure probability distribution of the safety system. In the following steps, the number of end states were used to perform the likelihood function for updating the prior distribution of the failure probability of the safety system which may result in the formation of the posterior function. And, consequently, the incident-probability of each end state can be deduced by using the Monte Carlo simulation. The results of our analysis show that the fire or poisoning-incidence probability proves bigger than the prior, whereas the overall safety performance of liquid ammonia tank tends to drop. The failure rates of the alarming system, automatic shutdown system and fire fighting system began to work more efficiently than before, that is to say, getting lower than before. With the safety systems failure rate becoming lower and lower, the uncertainty rate of the accidents has been reduced due to the use of Bayesian theory. Thus, it can be concluded that the dynamic risk analysis method proposed by us can help to improve the system＇s security status remarkably and enable the policymakers to take more effective urgent measures.